Skip to main content
SubZeroDev

Professional uncertainty since 2026.

Well… Why not?


All of It!

· 3 min read

Reddit asked:

How much AI should I use at work?

People had thoughts.

Many thoughts.

Responsible thoughts.

Nuanced thoughts.

Thoughts about craftsmanship.

Thoughts about productivity.

Thoughts about expertise.

Thoughts about whether agents make programming boring.

Thoughts about bug ratios.

Thoughts about design.

Thoughts about maintaining skill.

I read all this philosophy.

Then I typed:

All of it!

That was my contribution.

No framework.

No maturity model.

No AI adoption matrix.

No carefully balanced statement about preserving human agency while augmenting cognitive throughput.

All.

Of.

It.

Apparently This Requires Clarification​

I do not mean:

Let an AI randomly commit garbage to production while I nap under a desk.

I still review things.

I still design systems.

I still decide what I am trying to build.

I still notice when an abstraction is wrong.

I still know when the model has wandered into the weeds.

Unfortunately.

What I mean is simpler.

If a machine can remove friction from work I already know how to judge, why would I deliberately avoid using it?

Because somebody thinks typing every line manually is character building?

No.

I have built enough character.

I Am Not Protecting the Shovel​

Imagine somebody invents an excavator.

Then a group of professional hole diggers gather around and say:

Real digging requires a shovel.

Fine.

Use the shovel.

Enjoy the shovel.

Become one with the shovel.

I am using the fucking excavator.

I still need to know where the hole goes.

I still need to avoid the gas line.

I still need to know whether the building next door will collapse.

But I am not preserving repetitive labor because the repetitive labor used to be part of the job.

The Job Moved​

This is what actually changed for me.

The interesting part of software was never:

How many characters can I personally type?

The interesting part is:

What are we building?

Why?

Where are the boundaries?

What breaks?

What does the system actually guarantee?

How do the pieces fit?

What did we accidentally invent while solving the original problem?

AI moved more of my time toward those questions.

Good.

I was already trying to get there.

"But You'll Lose Your Skills"​

Maybe.

Some skills should probably decay.

I no longer remember every command-line flag I used twenty years ago.

Civilization survived.

I can look things up.

I can ask a model.

I can inspect the output.

I can still tell when something smells wrong.

That last part matters more to me than preserving perfect recall of syntax that a machine can produce in milliseconds.

The Funny Part​

The Reddit thread was full of people writing thoughtful paragraphs.

I am not mocking them.

Some of the comments were good.

Some raised legitimate tradeoffs.

But the original question was:

How much AI should I use?

And apparently my entire considered position had already compressed itself into:

All of it!

I laughed after posting it.

Because there it was again.

No engineered response.

No attempt to be clever.

Just the first thing that felt true.

And honestly?

Still does.